Fermentable by chemical hydrolysis of biomass

Joseph B. Binder and Ronald T. Raines1

Departments of Chemistry and , University of Wisconsin, Madison, WI 53706

Edited by Christopher R. Somerville, University of California, Berkeley, CA, and approved January 5, 2010 (received for review October 19, 2009)

Abundant plant biomass has the potential to become a sustainable 1948 (20, 21). In the United States, several related processes source of fuels and chemicals. Realizing this potential requires the using H2SO4 have been developed, typically with 80–90% conver- economical conversion of recalcitrant lignocellulose into useful sion of and into sugars (22–27). Neverthe- intermediates, such as sugars. We report a high-yielding chemical less, the hazards of handling concentrated and the process for the hydrolysis of biomass into . Add- complexities of recycling them have limited the adoption of this ing water gradually to a chloride ionic liquid-containing catalytic technology. leads to a nearly 90% yield of from cellulose and Less hazardous and more tractable cellulose solvents would 70–80% yield of sugars from untreated corn stover. -exclusion facilitate lignocellulose hydrolysis. Room-temperature ionic chromatography allows recovery of the ionic liquid and delivers sug- liquids, that is, salts with melting points near or below ambient ar feedstocks that support the vigorous growth of ethanologenic temperature, show promise as cellulose solvents for nonwoven microbes. This simple chemical process, which requires neither an fiber production (28) and chemical derivatization (29–31). Like edible plant nor a , could enable crude biomass to be the concentrated acids, ionic liquids containing chloride, , and sole source of carbon for a scalable biorefinery. other moderately basic anions disrupt the hydrogen bond net- work of cellulose and enable its dissolution (30–34). Although biofuel ∣ ∣ fermentation ∣ ionic liquid ∣ lignocellulose ionic liquids have been used to pretreat biomass for enzymatic hydrolysis (35–37), chemical deconstruction to produce glucose s the primary component of lignocellulosic biomass, cellulose in ionic liquids has produced only moderate yields (38–43) that CHEMISTRY Ais the most abundant organic compound on earth and has the contrast with the nearly quantitative yields of glucose attainable potential to be a renewable source for energy and chemicals. The from cellulose in concentrated acids and other cellulose sol- 11 estimated global annual production of biomass is 1 × 10 tons, vents (44). 21 sequestering 2 × 10 J (1, 2). For comparison, annual Noting that highly efficient hydrolyses employ much higher 20 production amounts to 2 × 10 J, whereas the technically recov- water concentrations than those used in ionic-liquid-phase hydro- 22 erable endowment of conventional crude oil is 2 × 10 J (1). lysis, we reasoned that water could play an important role in

Hence, in only one decade, Earth’s plants can renew in the form delivering high glucose yields. Our investigation of the reactivity SCIENCES of cellulose, hemicellulose, and lignin all of the energy stored as of glucose and cellulose in mixtures of an ionic liquid and water conventional crude oil. The challenge for scientists is to access confirmed this hypothesis and enabled a high-yielding process for APPLIED BIOLOGICAL these and convert them into fuels and building blocks the hydrolysis of cellulose and lignocellulosic biomass. Moreover, for civilization. this process generates easily recovered sugars that are superb Sugars are natural intermediates in the biological and chemical feedstocks for microbial growth and biocatalytic ethanol pro- – conversion of lignocellulosic biomass (3 11), but access to sugars duction. is hindered by the recalcitrance of plant cell walls (4, 12). The majority of glucose in lignocellulose is locked into highly crystal- Results — line cellulose polymers. Hemicellulose a branched of Cellulose Reactivity in Ionic Liquids. We began by investigating the — — glucose, xylose, and other sugars and lignin a complex aro- intrinsic reactivity of cellulose and sugars under acidic conditions — matic polymer encase the cellulose, fortifying and protecting in ionic liquids. First, we reacted cellulose with H2SO4 and HCl in the plant. Deriving sugars from this heterogeneous feedstock re- 1-ethyl-3-methylimidazolium chloride ([EMIM]Cl), an ionic quires both physical and chemical disruption. Enzymatic methods liquid that is an effective solvent for cellulose (39). We observed of saccharification are the most common, and use physical and the production of 5-hydroxymethylfurfural (HMF), but only mod- chemical pretreatment processes (13) followed by hydrolysis with erate yields of glucose (Table 1). HMF inhibits microbial growth to produce sugars. The proper combination of pre- and fermentation (45). treatment and for a given feedstock enables high yields The production of HMF at the expense of glucose suggested to of sugars from both hemicellulose and cellulose components us nascent glucose was being dehydrated to form HMF. To test (14, 15). Nonetheless, the costs of both pretreatment and en- this hypothesis, we reacted glucose in [EMIM]Cl containing vary- zymes [one-third of the cost of ethanol production from cellulose ing amounts of water (Fig. 2, Table S1 in the SI Appendix). In the (16)] and low rates of hydrolysis are potential drawbacks to absence of both acid and water, glucose was recovered intact. On enzymatic hydrolysis. the other hand, adding H2SO4 led to the rapid decay of glucose Chemistry provides an alternative means to hydrolyze biomass. into HMF and other products in ionic liquid containing little or As early as 1819, Braconnot demonstrated that linen dissolved in no water. Most significantly, we observed that increasing the concentrated H2SO4, diluted with water, and heated was trans- formed into a fermentable (17, 18). The concentrated acid plays a dual role in biomass hydrolysis. By disrupting its network Author contributions: J.B.B. and R.T.R. designed research; J.B.B. performed research; of intra- and interchain hydrogen bonds, strong acids decrystallize J.B.B. and R.T.R. analyzed data; J.B.B. and R.T.R. wrote the paper. cellulose and make it accessible to reagents (19); and by catalyz- The authors declare no conflict of interest. ing the hydrolysis of glycosidic bonds, strong acids cleave cellu- This article is a PNAS Direct Submission. lose and hemicellulose into sugars (Fig. 1) (4). Bergius took 1To whom correspondence should be addressed. E-mail: [email protected]. advantage of these attributes of HCl in the development of a This article contains supporting information online at www.pnas.org/cgi/content/full/ commercial process that operated in Germany from 1935 to 0912073107/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.0912073107 PNAS Early Edition ∣ 1of6 Downloaded by guest on September 29, 2021 Fig. 1. Hydrolysis reactions of cellulose and xylan. Chemical hydrolysis of cellulose and hemicellulose into monomeric sugars proceeds through oligomers and is accompanied by side reactions that form furans and other degradation products.

water content to 33 wt% in this same acidic enabled at 105 °C to allow hydrolysis of the cellulose into shorter, more nearly 90% of the glucose to remain intact after 1 h. soluble segments (Table 1). After a delay, additional water was Why does glucose degrade rapidly under nonaqueous condi- added to stabilize the glucose product. tions in [EMIM]Cl but not in the presence of a high concentra- The timing of water addition was critical for high glucose tion of water? Le Chatelier’s principle (46) tells us that water yields. When the reaction mixture was diluted to 33% water after disfavors dehydrative reactions, such as glucose oligomerization 5 min, cellulose precipitated, resulting in low yields. Delaying and conversion into HMF. Additionally, the highly nucleophilic dilution until after 10 min prevented cellulose precipitation, chloride anions of [EMIM]Cl coordinate strongly to carbohy- and gradually increasing the water content to 43% within drates (47, 48), accelerating acid-catalyzed dehydration reactions 60 min allowed glucose yields of nearly 90% in 2–4 h. These yields (49). High concentrations of water solvate chloride and thus are nearly twice as high as the previous best in ionic liquids and distract it from interacting with . approach those achieved through enzymatic hydrolysis. We also examined the effect of the time for dissolution of cellulose in the Effect of Water on Celluose Reactivity in Ionic Liquids. We suspected ionic liquid prior to hydrolysis, finding that 6 h was optimal that increasing the water concentration in an [EMIM]Cl reaction (Table S2 in the SI Appendix). Longer times probably led to in- mixture would enhance glucose yields from cellulose. We were, creased byproduct formation (50, 51), which was indicated by dis- however, aware that water precipitates cellulose from ionic coloration of the reaction mixture, whereas shorter times were liquids (32). Indeed, we observed that a 5 wt% solution of cellu- insufficient for complete of the cellulose. With this op- lose in [EMIM]Cl formed an intractable gel when the solution timized procedure, more concentrated cellulose (10 wt was diluted to 10 wt% water, making homogeneous hydrolysis %) could be hydrolyzed in high yields. Tiny cellulose fibers visible of cellulose in ionic liquid/water solutions impossible. Instead, in these reaction mixtures suggest that glucose yields were slightly we attempted to balance cellulose solubility and glucose stability lower due to incomplete cellulose breakdown prior to water ad- by adding water gradually during hydrolysis, expecting that cellu- dition. It is likely that cellulose is converted into a mixture of glu- lose solubility would increase as the reaction progresses. In these cose and soluble oligomers within the first 30–60 min of reaction, experiments, we chose to use HCl as the acid so as to match the and that these oligomers hydrolyze subsequently into glucose. anion with that of the ionic liquid. [EMIM]Cl containing 5 wt% Monitoring the production of glucose and (which cellulose was first treated with HCl and a small amount of water is a glucose dimer) during cellulose hydrolysis revealed that

Table 1. Hydrolysis of cellulose in [EMIM]Cl Water content, wt% Cellulose, wt% HCl, wt% 0′ 5′ 10′ 20′ 30′ 60′ Time, h Glucose yield, % HMF yield, % 5 20* 5 5 5 5 5 5 1 40 19 5 20555555 1 45 17 5 20 5 33 33 33 33 33 1 14 ND 229ND 5 205 5 202020201 31 ND 26419 35125 43630 5 205 5 202033331 40 ND 2847 38110 4778 5 205 5 202533331 44 ND 2867 38310 47713 5 205 5 202533431 38 ND 2855 3876 4897 10 10 5 5 20 25 33 43 3 71 ND Cellulose was reacted in [EMIM]Cl at 105 °C after its dissolution at 105 °C for 12 h. HCl loading is relative to cellulose mass. Water content is relative to the total mass of the reaction mixture and is shown at time points (minutes) following the initiation of hydrolysis. Yields are molar yields based on HPLC analysis, and are relative to the glucose monomers contained in the cellulose. ND, not determined. *H2SO4.

2of6 ∣ www.pnas.org/cgi/doi/10.1073/pnas.0912073107 Binder and Raines Downloaded by guest on September 29, 2021 Table 2. Hydrolysis of cellulose in ionic liquids Cellulose concentration, Glucose Ionic liquid wt% yield, % [EMIM]OAc 2 0 [EMIM]OAc 5 0 ½EMIMNO3 20 1,3-dimethylimidazolium 20 dimethylphosphate 1,3-dimethylimidazolium 50 dimethylphosphate [EMIM]Br 2 4 ½BMIMBF4 20 [EMIM]OTf 2 5 [BMIM]Cl 5 66 1-butyl-4-methylpyridinium chloride 5 73 1-ethylpyridinium chloride 5 69 1-ethyl-2,3-dimethylimidazolium 546 chloride Cellulose was reacted in ionic liquid for 3 h at 105 °C after mixing at 105 °C Fig. 2. Acid-catalyzed degradation of glucose in [EMIM]Cl. In acidic [EMIM] for 6 h. HCl loading was 20 wt% relative to cellulose weight. The water Cl, glucose (diamonds) disappears rapidly at 100 °C, forming HMF (squares) content of the reaction began at 5 wt% and was increased as follows: and other degradation products. Increased water content slows glucose loss. 20% (10 min), 25% (20 min), 33% (30 min), 43% (60 min). Yields are molar yields based on HPLC analysis and are relative to the glucose Reaction conditions: Glucose, 10 wt%; H2SO4, 4 wt% relative to glucose; H2O, 0–33 wt% relative to the total mass of the reaction mixture. Product levels monomers contained in the cellulose. are reproducible to within 4%. of glucose based on the xylan and cellulose content of the stover (Table 3). Dilution of the reaction mixture to 70% water caused cellobiose concentrations peaked at 1 h and decayed as glucose CHEMISTRY concentrations increased (Fig. 3). Experiments with other ionic precipitation of unhydrolyzed and lignin. These liquids indicate that chloride-based ionic liquids alone are able residues were then dissolved in [EMIM]Cl and subjected to an to balance cellulose solubility and hydrolytic activity through identical second-stage hydrolysis, which released additional strong interactions with cellulose coupled with weak basicity xylose and glucose, leaving behind lignin-containing solids. (Table 2). Combined, these two steps resulted in a 70% yield of glucose and a 79% yield of xylose using only simple chemical reagents. Hydrolysis of Crude Biomass. Complex and heterogeneous, lignocel- Given the well-documented versatility of ionic liquids as biomass lulosic biomass presents a more significant challenge for hydro- solvents, we anticipate that this process will be amenable to other SCIENCES lysis than does cellulose. In addition to intractable crystalline biomass sources, such as wood and grasses. The brown residue APPLIED BIOLOGICAL cellulose, lignocellulosic biomass such as corn stover includes remaining after two-stage corn stover hydrolysis likely includes protective hemicellulose and lignin, heterogeneous components lignin, insoluble polysaccharides, any insoluble hydrolysis that are major obstacles to many biomass hydrolysis processes byproducts, , and ash. We have not yet characterized (4, 12). Nonetheless, chloride ionic liquids are excellent solvents the lignin byproduct, but it is likely to resemble lignin produced for lignocellulosic biomass, suggesting that they would easily by other processes, e.g., Klason lignin (43, 52). disrupt these polymeric barriers. Moreover, preliminary experi- ments demonstrated that xylan, a hemicellulose, was readily hy- Sugar and Solvent Recovery. A practical biomass hydrolysis process drolyzed under our conditions, producing xylose in 77% yield. requires efficient means for sugar and solvent recovery. We found We attempted to extend our process for cellulose hydrolysis to that ion-exclusion chromatography enables separation of the crude lignocellulosic biomass. Untreated corn stover that had sugars and ionic liquid from the corn stover hydrolysis reaction been mixed with [EMIM]Cl was hydrolyzed with 10 wt% HCl mixture. In this technique, a mixture containing electrolyte and at 105 °C with the same water-dilution process used for pure cel- nonelectrolyte solutes is separated by passing it through a lulose. This process produced a 71% yield of xylose and 42% yield charged resin (53). Charged species, such as the ionic liquid, are excluded from the resin, while nonelectrolytes, such as sugars, are retained. Nonpolar species such as HMF and furfural are adsorbed more strongly than sugars, and elute later. Passing the corn stover hydrolyzate through a column of [EMIM]- exchanged Dowex® 50 resin allowed laboratory-scale separation of the ionic-liquid solvent from the sugars, with >95% recovery of the ionic liquid, 94% recovery of glucose, and 88% recovery of xylose. Additionally, this chromatographic step removed inhibi- tory compounds such as HMF and furfural. We suspect that ionic-liquid and sugar recovery are limited by the difficulties intrinsic in small-scale separations and would be improved upon scale-up.

Microbial Fermentation of Biomass Hydrolyzate. To support biocon- version, biomass hydrolyzate sugars must be free of contaminants that inhibit microbial growth and fermentation. We found that Fig. 3. Production of glucose, HMF, and cellobiose production during sugars derived from corn stover through our process are excellent cellulose hydrolysis under optimized reaction conditions. Product levels feedstocks for an ethanologenic bacterium and yeast. Whereas are reproducible to within 4%. wild-type Escherichia coli ferments a range of sugars into a

Binder and Raines PNAS Early Edition ∣ 3of6 Downloaded by guest on September 29, 2021 Table 3. Hydrolysis of untreated corn stover in [EMIM]Cl Water content, wt% Stover, wt% Stage HCl, wt% 0′ 5′ 10′ 20′ 30′ 60′ Time, h Glucose yield, % Xylose yield, % 5 1 20 5 5 20 25 33 43 2.5 42 71 2 2055202533433.0 28 8 Overall 70 79 10 1 1055202533433.5 19 74 2 1055202533433.0 47 1 Overall 66 75 10 1 1055202533431.0 17 60 1 1.5 21 73 1 2.0 25 80 1 2.5 27 82 2 1055202533431.5 37 5 2 2.0 42 5 2 2.5 44 5 2 3.0 48 5 Untreated corn stover was reacted in [EMIM]Cl at 105 °C after its dissolution at 105 °C for 6 h. HCl loading is relative to stover weight. Water content is relative to the total mass of the reaction mixture and is shown at time points (minutes) following the initiation of hydrolysis. Yields are molar yields based on HPLC analysis and are relative to the glucose and xylose monomers contained in the stover.

mixture of ethanol and organic acids, the engineered KO11 strain produces ethanol selectively (54). Serving as the sole carbon source, the hydrolyzate sugars from corn stover enabled aerobic growth of E. coli KO11 at a rate comparable to that of a control glucose/xylose mixture (Fig. 4A). Moreover, under oxygen- deficient conditions, E. coli KO11 produced a ð79 4Þ% yield of ethanol from stover hydrolyzate sugars and a ð76 3Þ% yield from pure glucose and xylose, demonstrating that sugars from our hydrolysis process can be converted readily into ethanol. Engineered bacteria show promise for biofuel production, but yeast fermentation predominates today (55, 56). Pichia stipitis, which has an innate ability to ferment xylose, is an especially viable candidate for bioconversion of lignocellulose-derived sugars (57–59). Corn stover hydrolyzate sugars are an excellent carbon source for the growth of this yeast (Fig. 4B), and P. stipitis efficiently converts them into ethanol. Fermenting xylose and glucose, the yeasts produced a ð70 2Þ% yield of ethanol from hydrolyzate and a ð72 1Þ% yield from pure sugars. Discussion We have demonstrated an efficient system for hy- drolysis as well as means to separate and ferment the resulting sugars. By balancing cellulose solubility and reactivity with water, we produce sugars from lignocellulosic biomass in yields that are severalfold greater than those achieved previously in ionic liquids and approach those of enzymatic hydrolysis. Furthermore, the hy- drolyzate products are readily converted into ethanol by micro- organisms. Together, these steps comprise an integrated process for chemical hydrolysis of biomass for biofuel production (Fig. 5). First, lignocellulosic biomass such as corn stover is decrystallized through mixing with [EMIM]Cl. With their defense against chemical assault breached by the ionic liquid, the hemicellulose and cellulose are hydrolyzed by treatment with HCl and water. The residual lignin and cellulose solids are subjected to a second hydrolysis, while the liquid hydrolyzate is separated through ion- exclusion chromatography. Ionic liquid recovered in the ion- exclusion step is stripped of water and recycled, while hydrolyzate sugars are fermented into fuels and other products. In comparison to extant enzymatic and chemical processes to biomass hydrolysis, ours has several attractive features. Like con- Fig. 4. Aerobic growth of ethanologenic microbes on corn stover hydroly- centrated acid processes, it uses inexpensive chemical catalysts zate and pure sugars as their sole carbon source. (A) Escherichia coli strain ⦁ ○ rather than enzymes and avoids an independent pretreatment KO11: hydrolyzate ( ), 5 replicates; sugars ( ), 20 replicates. (B) Pichia stipitis: hydrolyzate (⦁), 2 replicates; sugars (○), 6 replicates. Data are mean values step. Working in concert, [EMIM]Cl and HCl produce high sugar (SE) at 0.5 h intervals. Under anaerobic conditions, the cultures produced yields in hours at just 105 °C, whereas enzymatic hydrolysis can ethanol with the following yields based on sugar conversion (three replicates take days (16) and many pretreatment methods require tempera- each). Escherichia coli strain KO11: hydrolyzate, ð79 4Þ%; sugars, tures of 160–200 °C (13). Also, lignocellulose solubilization by the ð76 3Þ%. Pichia stipitis: hydrolyzate, ð70 2Þ%; sugars, ð72 1Þ%.

4of6 ∣ www.pnas.org/cgi/doi/10.1073/pnas.0912073107 Binder and Raines Downloaded by guest on September 29, 2021 Fig. 5. Integrated process for biofuel production using ionic-liquid biomass hydrolysis. Ionic-liquids solvents enable efficient biomass decrystallization and hydrolysis. Ion-exclusion chromatography separates the ionic liquids for recycling and the hydrolyzate sugars for fermentation. (Photograph copyright 2001, Department of Energy/National Renewable Energy Laboratory; credit, Jim Yost.)

ionic liquid allows processing at high concentrations, which can Our chemical hydrolysis method offers flexibility for an be a problem in enzymatic hydrolysis. On the other hand, our pro- integrated biomass conversion process. Because the ionic-liquid cess improves on typical acid hydrolysis methods by avoiding the solvent makes biomass polysaccharides readily accessible for use of hazardous concentrated acid. Using catalytic amounts of chemical reactions, this process is likely to be compatible with dilute acid removes the complexity and danger of recycling large a broad range of biomass feedstocks. Downstream, the sugars volumes of concentrated acid. The ionic liquid used in its place is produced by lignocellulose hydrolysis are flexible feedstocks likely to be far easier to handle. Despite this difference, our pro- for production of a nearly infinite range of fuels and chemicals. cess is similar to commercial processes using concentrated acid E. coli, which readily use the hydrolyzate sugars, have been hydrolysis (20, 27) and consequently can exploit proven engineered to produce not only fuel ethanol but also 1-butanol, engineering and equipment for facile scale-up, particularly for se- 2-butanol, branched , fatty acids, isoprenoids, and even parations and recycling (vide infra). hydrogen (63–65). Furthermore, the aqueous stream of sugars Still, important challenges remain for implementation of an can also be converted by catalytic processes into fuels (66) or che- ionic-liquid biomass hydrolysis process. Highly viscous bio- mical intermediates. In contrast with enzymatic hydrolysis reac- mass-ionic-liquid mixtures might require special handling, and tions that often require coupling to fermentation (simultaneous CHEMISTRY larger scale fermentation of hydrolyzate sugars might reveal saccharification and fermentation) to prevent product inhibition the presence of inhibitors not detected in our demonstration ex- (15, 16), this chemical process can be paired with any downstream periments. Moreover, the sugar concentration resulting from conversion. Finally, the lignin recovered from ionic-liquid bio- stover hydrolysis (about 1%) is too low for practical fermentation mass hydrolysis could be a valuable coproduct. After washing and is decreased even further during the chromatographic se- to remove ionic liquid, this residue could be burned to generate paration, leading to water-evaporation costs. Methods allowing process heat and electricity. The lignin residue from similar acid

a higher starting biomass loading and strategies to concentrate hydrolysis processes has a heating value similar to that of wood, SCIENCES rather than dilute the sugars during separation from the ionic making it a useful feed for a solid fuel boiler (62). Alternatively, liquid would overcome this problem. the lignin could be an excellent feedstock for high-value products APPLIED BIOLOGICAL Separations and ionic-liquid recycling could pose additional (43, 67). As a result, our process, which uses simple chemical re- challenges to commercialization. Our separation method— agents to overcome biomass recalcitrance and liberate valuable ion-exclusion chromatography—is complex, especially with acid, sugars, has the potential to underlie a biorefinery. ionic liquid, multiple sugars, and other soluble species in the Materials and Methods hydrolyzate. Nonetheless, this technique is already used in con- A description of general materials and methods, the hydrolysis of cellulose tinuous carbohydrate processing on a large scale, notably for the and xylan, the recovery of sugars and [EMIM]Cl, the reaction of glucose, ex- separation of glucose and in corn wet mills (60). The se- periments with nonchloride ionic liquids, microbial growth and fermenta- paration of electrolytes such as ionic liquids from sugars is likely tion, the economics of ionic liquid usage, and Tables S1 and S2 are to be easier than this industrial application, and is currently being provided in SI Text. scaled up by BlueFire Ethanol, Inc. with in two demonstration plants. Lessons from these facilities can aid in Representative Procedure for Hydrolysis of Corn Stover. Corn stover (26.7 mg, 54 μmol glucose units, 44 μmol xylose units) and [EMIM]Cl (502 mg) were the design of efficient processes based on ionic-liquid hydrolysis. mixed at 105 °C for 6 h. To this mixture was added aqueous HCl (1.66 M, Ionic liquids are expensive and will likely remain so in compar- 29 μL, equivalent to 5 mg conc. HCl), and the reaction mixture was stirred ison to other solvents, even as expanded use reduces their prices vigorously at 105 °C. After 10 min, deionized water (100 μL) was added with to near those of their raw materials [∼$2.50∕kg for dialkylimida- stirring, followed by additional aliquots at 20 min (50 μL), 30 min (75 μL), and zolium chlorides (61)]. Consequently, virtually complete solvent 60 min (125 μL). After a total reaction time of 2.5 h, the solution was diluted recovery will be required to make biomass processing with ionic with water (750 μL). Insoluble materials were removed by centrifugation, μ liquids economical. For instance, one target for the cost of ionic rinsed twice with water (200 L), and dried. The liquid products (2.046 g) were 2 0 ∕ 2 3 ∕ liquid lost in biomass processing might be $0.13∕L(¼$0.50∕gal) analyzed by HPLC ( . mg g glucose, 42% yield; . mg g xylose, 71% yield). The brown solids from the first hydrolysis were then heated with [EMIM]Cl ethanol, a near-term target for the cost of hydrolytic enzymes. An (306 mg) at 105 °C for 4.5 h. To this mixture was added aqueous HCl (1.66 M, optimized process using 1 kg of ionic liquid for each kilogram of 14.5 μL, equivalent to 2.5 mg conc. HCl), and the reaction mixure was vigor- biomass processed would then need 98% recovery of the ionic ously stirred at 105 °C. After 10 min, deionized water (50 μL) was added with liquid in each cycle (see SI Text). This level of solvent recovery stirring, followed by an additional 25 μL water at 20 min, 67.5 μL water at might be feasible in an optimized process, considering that re- 30 min, and 70 μL water at 60 min. After 3 h total reaction time, the solution ported sulfuric acid losses in a pilot-scale concentrated acid pro- was diluted with water (300 μL) and centrifuged to sediment insoluble cess are only about 38 g∕kg biomass, despite the much lower materials. The liquid products (770 mg) were analyzed by HPLC (3.56 mg∕g glucose, 28% yield; 0.7 mg∕g xylose, 8% yield). For the two-step incentive to recover sulfuric acid (62). We note, however, that process, the overall yield of glucose was 70% and the overall yield of xylose the accumulation of impurities in an ionic-liquid recycle loop was 79%. could be problematic. Future studies will need to address In other cases, aliquots of the reaction mixture were removed periodically this issue. for HPLC analysis.

Binder and Raines PNAS Early Edition ∣ 5of6 Downloaded by guest on September 29, 2021 ACKNOWLEDGMENTS. We are grateful to J.E. Holladay for helpful conversa- W. Jeffries and T.M. Long for assistance with yeast fermentation. This work tions on biomass chemistry in ionic liquids; B.R. Caes for contributive discus- was supported by the Department of Energy through the Great Lakes Bioe- sions; J.J. Blank and A.V. Cefali for experimental assistance; W.D. Marner, G.N. nergy Research Center. J.B.B. was supported by an National Science Founda- Phillips, and T.J. Rutkoski for assistance with bacterial growth studies; and T. tion Graduate Research Fellowship.

1. US National Petroleum Council (2007) Facing the Hard Truths about Energy (US Na- 36. Kilpelainen I, et al. (2007) Dissolution of wood in ionic liquids. J Agric Food Chem tional Petroleum Council, Washington, DC), pp 110, 156–158. 55:9142–9148. 2. Smeets EMW, Faaij APC, Lewandowski IM, Turkenburg WC (2007) A bottom-up 37. Singh S, Simmons BA, Vogel KP (2009) Visualization of biomass solubilization and assessment and review of global bio-energy potentials to 2050. Prog Energ Combust cellulose regeneration during ionic liquid pretreatment of switchgrass. Biotechnol 33:56–106. Bioeng 104:68–75. 3. Hahn-Hägerdal B, Galbe M, Gorwa-Grauslund MF, Lidén G, Zacchi G (2006) Bio-ethanol 38. Fanselow M, Holbrey JD, Seddon KR (2007) Eur Patent Appl 1,860,201. —The fuel of tomorrow from the residues of today. Trends Biotechnol 24:549–556. 39. Li C, Zhao ZK (2007) Efficient acid-catalyzed hydrolysis of cellulose in ionic liquid. Adv 4. (2005) Polysaccharides: Structural Diversity and Functional Versatility, ed Dumitriu S Synth Catal 349:1847–1850. (Marcel Dekker, New York), 2nd Ed. 40. Li C, Wang Q, Zhao ZK (2008) Acid in ionic liquid: An efficient system for hydrolysis of – 5. Peters D (2007) Raw materials. Adv Biochem Eng Biotechnol 105:1 30. lignocellulose. Green Chem 10:177–182. 6. Lange J-P (2007) Lignocellulose conversion: An introduction to chemistry, process and 41. Rinaldi R, Palkovits R, Schüth F (2008) Depolymerization of cellulose using solid – economics. Biofuel Bioprod Bior 1:39 48. catalysts in ionic liquids. Angew Chem Int Edit 47:8047–8050. 7. Stephanopoulos G (2007) Challenges in engineering microbes for biofuels production. 42. Vanoye L, Fanselow M, Holbrey JD, Atkins MP, Seddon KR (2009) Kinetic model for the – Science 315:801 804. hydrolysis of lignocellulosic biomass in the ionic liquid, 1-ethyl-3-methyl-imidazolium 8. Chheda JN, Huber GW, Dumesic JA (2007) Liquid-phase catalytic processing of chloride. Green Chem 11:390–396. biomass-derived oxygenated hydrocarbons to fuels and chemicals. Angew Chem Int 43. Sievers C, et al. (2009) Ionic-liquid-phase hydrolysis of pine wood. Ind Eng Chem Res Edit 46:7164–7183. 48:1277–1286. 9. Van Haveren J, Scott EL, Sanders J (2008) Bulk chemicals from biomass. Biofuel Bioprod 44. Cao NJ, Xu Q, Chen LF (1995) Acid hydrolysis of cellulose in solution. Appl Bior 2:41–57. Biochem Biotechnol 51–52:21–28. 10. Christensen CH, Rass-Hansen J, Marsden CC, Taarning E, Egeblad K (2008) The renew- 45. Almeida JRM, Betilsson M, Gorwa-Grauslund MF, Gorsich S, Liden G (2009) Metabolic able chemicals industry. ChemSusChem 1:283–289. effects of furaldehydes and impacts on biotechnological processes. Appl Microbiol 11. Binder JB, Raines RT (2009) Simple chemical transformation of lignocellulosic biomass Biotechnol 82:625–638. into furans for fuels and chemicals. J Am Chem Soc 131:1979–1985. 46. Le Chatelier HL (1884) On a general expression of the laws of 12. Himmel ME, et al. (2007) Biomass recalcitrance: Engineering plants and enzymes for – biofuels production. Science 315:804–807. (Translated from French). CR Hebd Seances Acad Sci 99:786 789. 13. Mosier N, et al. (2005) Features of promising technologies for pretreatment of 47. Remsing RC, Swatloski RP, Rogers RD, Moyna G (2006) Mechanism of cellulose dissolu- 13 35∕37 lignocellulosic biomass. Bioresource Technol 96:673–686. tion in the ionic liquid 1-n-butyl-3-methylimidazolium chloride: A C and Cl NMR – 14. Wyman CE, et al. (2005) Comparative sugar recovery data from laboratory scale relaxation study on model systems. Chem Commun 1271 1273. N0 application of leading pretreatment technologies to corn stover. Bioresource Technol 48. Remsing RC, et al. (2008) Solvation of carbohydrates in -dialkylimidazolium ionic – 96:2026–2032. liquids: A multinuclear NMR spectroscopy study. J Phys Chem B 112:11071 11078. 15. Lau MW, Dale BE (2009) production from AFEX-treated corn 49. Mascal M, Nikitin EB (2008) Direct, high-yield conversion of cellulose into biofuel. stover using Saccharomyces cerevisiae 424A(LNH-ST). Proc Natl Acad Sci USA Angew Chem Int Edit 47:7924–7926. 106:1368–1373. 50. Ebner G, Schiehser S, Potthast A, Rosenau T (2008) Side reaction of cellulose with com- 16. Aden A, et al. (2002) Lignocellulosic Biomass to Ethanol Process Design and Economics mon 1-alkyl-3-methylimidazolium-based ionic liquids. Tetrahedron Lett 49:7322–7324. Utilizing Co-Current Dilute Acid Prehydrolysis and Enzymatic Hydrolysis for Corn 51. Vitz J, Erdmenger T, Hänsch C, Schubert US (2009) Extended dissolution studies of cel- Stover (National Renewable Energy Laboratory, Department of Energy, Golden, CO). lulose in imidazolium based ionic liquids. Green Chem 11:417–424. 17. Braconnot H (1819) Memoir on the conversion of wood particles in rubber, in sugar, 52. Hatfield RD, Jung HJG, Ralph J, Buxton DR, Weimer PJ (1994) A comparison of the and in a special natural acid, by means of sulfuric acid; conversion of the same woody insoluble residues produced by the Klason lignin and acid detergent lignin procedures. substance in ulmin by potash (Translated from French). Ann Chim Phys 12:172–195. J Sci Food Agr 65:51–58. 18. Stern AL (1895) X. Contributions to the chemistry of cellulose. I. Celluose-sulphuric 53. Asher DR (1956) Sugar purification by ion exclusion. Ind Eng Chem 48:1465–1466. acid, and the products of its hydrolysis. J Chem Soc, Trans 67:74–90. 54. Dien BS, Cotta MA, Jeffries TW (2003) Bacteria engineered for fuel ethanol produc- 19. Xiang Q, Lee YY, Pettersson PO, Torget RW (2003) Heterogeneous aspects of acid tion: Current status. Appl Microbiol Biotechnol 63:258–266. hydrolysis of α-cellulose. Appl Biochem Biotechnol 105:505–514. 55. Ho NWY, Chen Z, Brainard AP, Sedlak M (1999) Successful design and development of 20. Bergius F (1937) Conversion of wood to carbohydrates. Ind Eng Chem 29:247–253. genetically engineered Saccharomyces yeasts for effective cofermentation of glucose 21. Schoenemann K (1958) The perfecting of wood hydrolysis in the Rheinau process. and xylose from cellulosic biomass to fuel ethanol. Adv Biochem Eng Biotechnol – Chim Ind (Paris) 80:140 150. 65:163–192. 22. Dunning JW, Lathrop EC (1945) The saccharification of agricultural residues: A contin- 56. Van Maris AJA, et al. (2007) Development of efficient xylose fermentation in Sacchar- – uous process. Ind Eng Chem 37:24 29. omyces cerevisiae: Xylose isomerase as a key component. Adv Biochem Eng Biotechnol 23. Tsao GT, Ladisch MR, Voloch M, Bienkowski PR (1982) Production of ethanol and 108:179–204. – chemicals from cellulosic materials. Process Biochem 17:34 38. 57. Jeffries TW (1985) Emerging technology for fermenting D-xylose. Trends Biotechnol ’ 24. Wright JD, D Agincourt CG (1984) Evaluation of Sulfuric Acid Hydrolysis Processes for 3:208–212. Fuel Production (Solar Energy Research Institute, Department of Energy, Gold- 58. Agbogbo FK, Haagensen FD, Milam D, Wenger KS (2008) Fermentation of acid- en, CO). pretreated corn stover to ethanol without detoxification using Pichia stipitis. Appl Bio- 25. Wright JD, Power AJ (1987) Comparative technical evaluation of acid hydrolysis chem Biotech 145:53–58. processes for conversion of cellulose to alcohol. Energy from Biomass and Wastes, 59. Agbogbo FK, Coward-Kelly G (2008) Cellulosic ethanol production using the naturally (Institute of Gas Technology, Chicago, IL), 10, pp 949–971. occurring xylose-fermenting yeast, Pichia stipitis. Biotech Lett 30:1515–1524. 26. Farina GE, Barrier JW, Forsythe ML (1988) Fuel alcohol production from agricultural 60. Colonna WJ, Clarke MA, Godshall MA, Cleary M, White JS (2004) Kirk-Othmer Ency- lignocellulosic feedstocks. Energ Source Part A 10:231–237. clopedia of Chemical Technology, ed Kroschwitz JI (Wiley-Interscience, New York) 27. Farone WA, Cuzens JE (1998) US Patent 5,726,046. p 485. 28. Hermanutz F, Meister F, Uerdingen E (2006) New developments in the manufacture of 61. Reddy RG (2006) Ionic liquids: How well do we know them? J Phase Equilib Diff cellulose fibers with ionic liquids. Chem Fibers Int 56:342–344. – 29. Graenacher C (1934) US Patent 1,943,176. 27:210 211. 30. Zhu S, et al. (2006) Dissolution of cellulose with ionic liquids and its application: A mini- 62. Cuzens JC, Miller JR (1997) Acid hydrolysis of bagasse for ethanol production. Renew – review. Green Chem 8:325–327. Energ 10:285 290. 31. El Seoud OA, Koschella A, Fidale LC, Dorn S, Heinse T (2007) Applications of ionic 63. Maeda T, Sanchez-Torres V, Wood TK (2007) Enhanced hydrogen production from liquids in carbohydrate chemistry: A window of opportunities. Biomacromolecules glucose by metabolically engineered Escherichia coli. Appl Microbiol Biotechnol – 8:2629–2647. 77:879 890. 32. Swatloski RP, Spear SK, Holbrey JD, Rogers RD (2002) Dissolution of cellose with ionic 64. Atsumi S, Hanai T, Liao JC (2008) Non-fermentative pathways for synthesis of liquids. J Am Chem Soc 124:4974–4975. branched-chain higher alcohols as biofuels. Nature 451:86–89. 33. Mikkola J-P, et al. (2007) Ultrasound enhancement of cellulose processing in ionic 65. Fortman JL, et al. (2008) Biofuel alternatives to ethanol: Pumping the microbial well. liquids: From dissolution towards functionalization. Green Chem 9:1229–1237. Trends Biotechnol 26:375–381. 34. Pinkert A, Marsh KN, Pang S, Staiger MP (2009) Ionic liquids and their interaction with 66. Kunkes EL, et al. (2008) Catalytic conversion of biomass to monofunctional hydrocar- cellulose. Chem Rev 109:6712–6728. bons and targeted liquid-fuel classes. Science 322:417–421. 35. Dadi AP, Varanasi S, Schall CA (2006) Enhancement of cellulose saccharification kinetics 67. Lora JH, Glasser WG (2002) Recent industrial applications of lignin: A sustainable using an ionic liquid pretreatment step. Biotechnol Bioeng 95:904–910. alternative to nonrenewable materials. J Polym Environ 10:39–48.

6of6 ∣ www.pnas.org/cgi/doi/10.1073/pnas.0912073107 Binder and Raines Downloaded by guest on September 29, 2021